See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/287164374

Fetal Heart Rate Monitoring Based on Adaptive Noise Cancellation and Maternal
QRS Removal Window

Article · January 2009

CITATIONS READS

20 353

4 authors, including:

Mohd Alauddin Mohd Ali Kok Beng Gan

Universiti Kebangsaan Malaysia Universiti Kebangsaan Malaysia
197 PUBLICATIONS   3,018 CITATIONS    99 PUBLICATIONS   636 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

EEG signal processing View project

Biosensor View project

All content following this page was uploaded by Mohd Alauddin Mohd Ali on 15 July 2014.

The user has requested enhancement of the downloaded file.

European Journal of Scientific Research
ISSN 1450-216X Vol.27 No.4 (2009), pp.565-575
© EuroJournals Publishing, Inc. 2009
http://www.eurojournals.com/ejsr.htm

Fetal Heart Rate Monitoring Based on Adaptive Noise

Cancellation and Maternal QRS Removal Window

M. Sheikh M. Algunaidi
Student Member IEEE Department of Electrical, Electronic & Systems Engineering
Faculty of Engineering and Built
Environment, University Kebangsaan Malaysia (UKM), Bangi, Malaysia
E-mail: malgun@vlsi.eng.ukm.my

M. A. Mohd Ali
Member IEEE, Department of Electrical, Electronic & Systems Engineering
Faculty of Engineering and Built
Environment, University Kebangsaan Malaysia (UKM, Bangi, Malaysia
E-mail: mama@vlsi.eng.ukm.my

K. B. Gan
Member IEEE, Department of Electrical, Electronic & Systems Engineering
Faculty of Engineering and Built
Environment, University Kebangsaan Malaysia (UKM), Bangi, Malaysia

E. Zahedi
Member IEEE, School of Electrical Engineering, SHARIF University of Technology
Tehran, Iran

Abstract

In this paper a new method to extract the fetal signal from the abdominal
electrocardiogram (ECG) is presented. A three-stage method for fetal heart rate detection
from abdominal ECG recordings is proposed. After preprocessing, adaptive noise
cancellation (ANC) is used to extract the fetal ECG. Then in the third stage maternal QRS
complex removal window is applied to eliminate or scale down the maternal residual peaks.
The method is validated using 30 recorded data and compared with another three stage
method using independent component analysis (ICA) for the fetal ECG extraction. The
average sensitivity and average positive predictivity of the ANC based method is 85.8 %
and 67.6 % respectivly compared to 74.4% and 64.1% of the ICA based method. These
show that the ANC based method was more successful in detecting the FHR than ICA.

Keywords: Adaptive noise canceller, independent component analysis, fetal heart rate
monitoring and QRS Removal Window.
Fetal Heart Rate Monitoring Based on Adaptive Noise Cancellation and Maternal
QRS Removal Window 566

1. Introduction
Fetal heart rate (FHR) monitoring is one of the methodologies to test fetal well being and diagnose for
possible abnormalities. Fetal monitoring throughout the pregnancy enables the clinician to diagnose
and recognize the pathologic condition especially asphyxia [1].
Although Doppler ultrasound device is currently used for FHR monitoring, it is not suitable for
long term monitoring due to its sensitivity to movement and its safety for long term exposure has yet to
be established [2]. Besides ultrasound, non-invasive electrocardiography has been used to obtain
valuable clinical information about the fetal well being during pregnancy. The extraction of the fetal
electrocardiogram (ECG) can be carried out via skin electrodes attached to the maternal abdomen.
However, the abdominal ECG (AECG) is always corrupted with power line interference, maternal
ECG (MECG) and electromyogram where its variability is influenced by the gestational age, position
of the electrodes and the skin impedance [3]. Therefore, appropriate signal processing techniques are
required to reveal the fetal ECG (FECG) from the AECG.
Various research efforts have been proposed to extract the FECG from the AECG such as
adaptive filtering [4], correlation techniques [5], blind source separation [6] and a combination of
wavelet analysis and blind source separation methods [7]. FHR can be calculated by determining the
R-R intervals from the extracted FECG. However, the extracted FECG is still corrupted by the residual
peaks of MECG (especially its QRS complexes) hence the FECG detection remains difficult.
In this paper, an adaptive noise canceller (ANC) is proposed to extract the FECG from the
AECG. A QRS removal window (a window for removing the maternal QRS (MQRS) complex),
algorithm is developed to eliminate the MECG residual peaks in the extracted FECG. The performance
of the proposed algorithm is evaluated and compared with the well-known independent component
analysis (ICA) algorithm by using recorded data from the Universiti Kebangsaan Malaysia Medical
Center (PPUKM).

2. Methodology
2.1. Data Acquisition
AECG signals were recorded from 30 healthy pregnant women (at 35 to 38 weeks of gestation), most
of which are corrupted with different levels of noises, using the lead system as shown in Fig. 1. The
experimental protocol was approved by the PPUKM Research and Ethical Committee prior to
commencement of the study and informed consents were obtained from all subjects.

Figure 1: Locations of the abdominal electrodes.

com

P1 BIOPAC- MP
100A PC
P2
p4
P5
P3

The AECG signals, X(n) = [ X 1 (n ), X 2 (n ),..., X p (n )]T where n denotes a discrete-time index,
and T is the transpose operator, were simultaneously recorded from maternal abdomen using six
567 M. Sheikh M. Algunaidi, M. A. Mohd Ali, K. B. Gan and E. Zahedi

electrodes (five electrodes, p∈ [l, 2, 3, 4, 5], with a single common) using high gain amplifiers
(BIOPAC- MP 100A). The AECG signals were digitized at 1000 Hz with 12 bit resolution. The total
recording time during each session was about one minute.
Electrode p1 is located in such a way that only MECG signals are acquired while the electrodes
p2, p3, p4 and p5 acquired the mixture of MECG and FECG. Therefore, X 1 (n ) is defined as the
reference input and X 2 (n ) and X 3 (n ) are the primary input signals to the adaptive filter. As for ICA,
four of the acquired AECG signals, X(n) = [ X 2 (n ), X 3 (n ),..., X p (n )] p∈ [2, 3, 4, 5], are fed into the ICA
algorithm as X 1 (n ) contains only MECG signal

2.2. Algorithms
The block diagram of the proposed algorithm is shown in Figure (2). It consists of the pre-processing
stage, FECG extraction using ANC or ICA and the MQRS removal window.

Figure 2: The block diagram of the proposed algorithm.

Recorded ECG signals

Pre-processing functions

FECG extraction using ANC or ICA

MQRS Removal

FECG SIGNAL

2.2.1. Preprocessing Stage

The preprocessing stage consists of the removal of the DC signal, baseline wander and the power line
interference. Each observation signal is made zero mean by subtracting its mean as follows:
X (n) = X (n) − mean( X (n)) (1)
Baseline wander is caused by the patient's breathing or movements during recording. The
frequency of the baseline wander due to breathing is in the range of 1 Hz and the EMG noise (artifacts
of muscular contractions) is characterized by relatively high frequency noise, hence the recorded
signals were filtered by a FIR band-pass filter with cut-off frequencies at 4 Hz and 90 Hz.
The power line interference consists of 50 Hz sine wave and its harmonics. A notch filter
centered at 50 Hz is used to eliminate this interference. An example of an AECG signal and the pre-
processed signal are shown in Fig. 3.
Fetal Heart Rate Monitoring Based on Adaptive Noise Cancellation and Maternal
QRS Removal Window 568

Figure 3: (a) AECG signal and (b) pre-processed signal

(a)

(b)

2.2.2. FECG Extraction Technique

Two algorithms namely ANC and ICA have been implemented in this work to evaluate their
performance for FECG extraction.

A. Adaptive Noise Canceller

ANC is a method of estimating a signal (FECGp), contaminated by additive noise, MECG p , p∈ [2,3]
with the primary input to the ANC becoming Xp using a reference input, MECG1 ≅ X 1 [8]. The noise
MECG1 is uncorrelated with the FECGp but correlated in some unknown way with the noise
MECG p as shown in Figure (4). The X1 and Xp are the signals acquired from the maternal abdomen.

Figure 4: Adaptive noise canceller system

MECG p
X p = MECG p + FECG p
p∈ [2,3]
e
FECG p + +

y
MECG1 ≅ X 1 Adaptive filter

The noise MECG1 is filtered to produce an output y that is as close a replica as possible of
MECG p . This output is subtracted from the primary input X p to produce the system output
e = MECG p + FECG p − y (2)
Where y is the output of the adaptive filter Squaring both sides of Equation 2, we obtain
e 2 = FECG p2 + ( MECG p − y ) 2 + 2 FECG p ( MECG p − y ) (3)
Applying expectations on both sides of Equation 3, we get
569 M. Sheikh M. Algunaidi, M. A. Mohd Ali, K. B. Gan and E. Zahedi

E[e 2 ] = E[ FECG p2 ] + E[( MECG p − y ) 2 ] + 2 E[ FECG p ( MECG p − y )] (4)

As FECG is uncorrelated neither with MECG p nor with y then 2 E[ FECG p ( MECG p − y )] = 0 .
Finally, we obtain
E[e 2 ] = E[ FECG p2 ] + E[( MECG p − y ) 2 ] (5)
The goal of the adaptive filter is to minimize the mean square error (MSE) of
E[ MECG p − y ] = 0 This can be obtained iteratively, to give the optimal solution when y = MECG p .

B. Independent Component Analysis

ICA is a method to find underlying factors or components from multivariate (multidimensional)
statistical data. It looks for components that are both statistically independent and non-gaussian.
Although an excellent review has been given by Cichocki & Amari [9], a brief description is given
here.
Given a set of p mixed signals X(n) = [X1(n), X2(n), …, Xp(n)]T which are linear mixed with q
(p> q) unknown mutually statistically independent, zero-mean source signals S(n) = [s1(n), s2(n),…
,sq(n)]T and noise contaminated. This can be written as
q
X i (n) = ∑ Aij s j (n) + g ci (n) , i = 1,2,…,p (6)
j =1

or in the matrix notation

X = AS + g c (7)
where X = X(n) is the vector of sensor signals, S = S(n) is the source signal vector, gc = [gc(n), gc
(n),…, gcp(n)]T is the additive noise vector, A is an unknown p × q mixing matrix and n is the discrete-
time index. The noise vector, gc is assumed Gaussian and independent.
The mixing matrix A is determined by the body geometry and conductivity, as well as the
electrode-source relative positions [12]. Criteria based on maximization of non-gaussianity [13],
maximum likelihood, minimization of mutual information [14], tensorial methods [15] and non-linear
decorrelation [16] may be used to estimate the mixing matrix A and the source signal vector S.
In the noise-free model, gc = 0, the identification of the mixing matrix A and the sources signal,
S can be estimated if the sources are independent and non-Gaussian, and the number of sensors is equal
or larger than the number of independent sources to be estimated. However, a noisy estimates of the
sources signal may obtain, S = A -1 (X − g c ) , if gc ≠ 0. Therefore, pre-processing before applying ICA
may improve the performance of the ICA.
In this paper, Second Order Blind Identification (SOBI) algorithm has been utilized to extract
the FECG from the AECG. SOBI is a robust algorithm to separate the noises especially EMG and
electrode artifacts.

2.2.3. Maternal QRS Removal

In the post processing stage two steps are implemented which are the MQRS removal to eliminate the
maternal residual peaks, and finally a 1 Hz notch filter to attenuate the residual baseline wander in the
FECG.
MQRS signal is captured within a window which is defined by taking 50 samples before and
after every peak found in the input signal X1 with the condition as shown Fig. 5.
Fetal Heart Rate Monitoring Based on Adaptive Noise Cancellation and Maternal
QRS Removal Window 570

Figure 5: Flow Chart of the MQRS Removal Window

i=0
h=∞

i = i+1

h=i

X1(i)> X1 (i ± 1)
yes

no

no X1(i-s) = 0
i>h
yes
no yes
i > h + 2s

The QRS complex interval is given by

k = (i – s) : (i + s); (8)
where i is the sample index and s =50 is the number of samples. All samples that do not fall within this
window will be zero padded as shown in Figure 6(b).
The MQRS window of Figure6(b) is used to eliminate or scale down the maternal residual
peaks from the extracted FECG. A small amount of baseline wander was observed at this extracted
signal. Therefore, a notch filter centered at 1 Hz is used to attenuate this baseline wander.
571 M. Sheikh M. Algunaidi, M. A. Mohd Ali, K. B. Gan and E. Zahedi
Figure 6(a): Pre-processed AECG signal X1 and (b) MQRS interval definition.

(a)

(b)

2.3. Evaluation
The proposed algorithms have been implemented in Matlab codes using Matlab-7.4 (The Math-works
Inc.). The performances of the algorithms were then evaluated based on their sensitivities and positive
predictivities [15], when applied to AECG signal acquired from the PPUKM. The sensitivity is the
fraction of real events that are correctly detected and it is defined by,
TP
Se = (9)
TP + FN
The Positive Predictivity is the fraction of detections that are real events and it is defined by,
TP
+P= (10)
TP + FP
where FN (False Negatives) denotes the number of missed detections, FP (False Positives)
represents the number of extra detections and TP (True Positives) is the number of correctly detected
QRS complexes.

3. Results and Discussions

3.1. Adaptive noise canceller extraction technique
Examples of the extracted FECG using ANC, maternal QRS and FECG signal after applying maternal
QRS removal window are shown in Figure (7). It is noted that the maternal residual peaks are still
observed after the ANC and only eliminated after applying the maternal QRS removal window. After
maternal residual peaks have been eliminated from the extracted FECG signal, a small amount of
baseline wander has been observed. Therefore, notch filter centered at 1 Hz is adequate to attenuate
this baseline wander.
Fetal Heart Rate Monitoring Based on Adaptive Noise Cancellation and Maternal
QRS Removal Window 572

Figure 7: (a) Extracted FECG using ANC, (b) MQRS and (c) FECG signal after applying MQRS removal
window.

3.2. Independent Component Analysis Extraction Technique

Figure (8)shows examples of the result using ICA. The maternal residual peaks are still observed in the
extracted FECG signal using ICA. The QRS Removal Window was then applied to remove the
maternal residual peak from the FECG. After maternal residual peaks have been eliminated from the
extracted FECG signal, a small amount of baseline wander has been observed. Therefore, notch filter
centered at 1 Hz is adequate to attenuate this baseline wander.
573 M. Sheikh M. Algunaidi, M. A. Mohd Ali, K. B. Gan and E. Zahedi
Figure 8: (a) Extracted FECG using ICA, (b) MQRS window and (c) FECG signal after applying MQRS
removal window

(a)

(b)

(c)

With this improvement on the extracted signal, the performance of the FECG extraction
techniques (ANC and ICA) is compared in terms of FHR detection. The FHR is calculated from the
RR interval after peak detection [16].

3.3. Performance Evaluation

In this section, the ANC and ICA methods are evaluated using sensitivity and positive predictivity. The
effect of the lead position in the primary input of the ANC is also evaluated.
Table 1 shows the performance using ICA and ANC based methods at signal extraction stage.
The average sensitivity of the ANC based method is 85.5 % (X2 as primary signal) as compared to
74.4% of the ICA based method. The average positive predictivity of the ANC based method is 67.6%
(X2 as primary signal) as compared with that of the ICA based method which is 64.1%. It shows that
the ANC based method was more successful in detecting the FHR than ICA. The QRS Removal
window was employed to improve this detection.

Table 1: Performance of ICA and ANC based method

ANC method
ICA method
Weeks No Signals X2 as primary signal X3 as primary signal
Se (%) + P (%) Se (%) + P (%) Se (%) + P (%)
35 2 79.0 54.5 77.5 53.1 66.6 48.8
36 13 88.4 77.8 86.2 70.4 74.8 72.2
37 6 91.1 68.9 78.4 67.9 79.9 70.1
38 9 84.8 69.6 87.9 75.5 76.4 65.4
85.5(%) 67.6(%) 82.5(%) 66.7(%) 74.4(%) 64.1(%)
Fetal Heart Rate Monitoring Based on Adaptive Noise Cancellation and Maternal
QRS Removal Window 574

With the availability of the multi-lead system for the ICA used in this work, it is possible to
evaluate the lead position for the primary signal of the ANC that gives optimum results. Hence a
comparison is made between X 3 as the primary signal and X 2 as also shown in Table 1.
The average sensitivity of the proposed algorithm from the primary signal X 2 is 85.8 % as
compared to 82.5 % with primary signal X 3 . Also the average positive predictivity of the proposed
algorithm is 67.6% with primary signal X 2 as compared to X 3 with 66.7%.
The performance of the algorithm was better than ICA for both locations, although electrode
location p2 (associated with X2) is better than p3. This shows that the location of the electrode plays an
important role in FHR detection.

4. Conclusion
The proposed algorithm (ANC with the QRS Removal Window) has been demonstrated to have better
performance to extract the fetal signal. This method can use only two leads and a common. By using
the MQRS removal window it is shown that it is possible to control the amplitude of the maternal QRS
complex in the extracted signal or eliminate it. This facilitates detection of the fetal peaks and therefore
the determination of FHR.
The limitation of the proposed algorithm is that only signals which acquired later than 35
gestation weeks are tested. Farther improvement is required to implement the algorithm on ECG signal
earlier than 35 gestation week.
Current work is in progress towards realizing an online FHR detection using 24 bit high
resolution multi-channel bio-amplifier and finally the proposed algorithm will be fully tested in the
clinical environment.

Acknowledgement
The authors would like to thank the Ministry of Science, Technology and Innovation, Malaysia; for
supporting this work under the Science Fund Grant 03-01-02-SF0255.
The authors would like also to express their gratitude to Dr. Farshid Soheili for providing the
clinical data for this paper and especially Professor Dr. Muhamad Abdul Jamil M. Yassin and
Associate Professor Dr. Shuhaila Ahmad for their assistance in collecting the clinical data.
 575 M. Sheikh M. Algunaidi, M. A. Mohd Ali, K. B. Gan and E. Zahedi

References
[1] R.K. Freeman; T. J. Garite; M. P. Nageotte, "Fetal Heart Rate Monitoring", chpt.1, pp. 1–4,
2003. Lippincott Williams & Wilkins ( 2003).
[2] G. M. Friesen, et al. "A Comparison of the Noise Sensitivity of Nine QRS Detection
Algorithms." IEEE Trans. Biomed. Engineering 37: 85–98 (1990).
[3] R.C. Goodlin, "History of fetal monitoring", Am.J.Obstet,Gynecol 133, 323-352 (1989).
[4] E. R. Ferrara and B. Widrow, "Fetal electrocardiogram enhancement by time-sequenced
adaptive filtering", IEEE Trans. Biomed. Eng. 29, 458–460 (1982).
[5] S. Abboud, A. Alaluf, S. Einav, and D. Sadeh, "Real time abdominal fetal ECG recording using
hardware correlator", Comput. Biol. Med., 22, 32–335 (1992).
[6] L. De Lathauwer, B. De Moor, and J. Vandewalle, "Fetal electrocardiogram extraction by
source subspace separation", in Proc. IEEE SP/ATHOS Workshop HOS, 134–138 (1995).
[7] J.G. Maria, and C.A. Jonathon, "Fetal Electrocardiogram Extraction by Sequential Source
Separation in the Wavelet Domain", IEEE Trans Biomed Eng, 52, 390-400 (2005). B.
Azzerboni, F.L. Foresta, N. Mammone, and F.C. Morabito, "A New Approach Based on
Wavelet-ICA Algorithms for Fetal Electrocardiogram Extraction", in Proc. 13th
EuropeanSymposium of Artificial Neural Networks, 27-29 (2005).
[8] B.Widrow, J.R., Jr.Glover, J.M.McCool, J. Kaunitz, C.S. Williams, R.H. Hearn, J.R. Zeidler,
Jr. E. Dong, R.C.Goodlin, "Adaptive noise cancelling: Principles and applications",
Proceedings of the IEEE 63, 1692 - 1716 (1975).
[9] A. Cichocki & S. Amari, "Adaptive Blind Signal and Image Processing", pp.157-175, Wiley
(2002).
[10] J. Vanderschoot, D. Callaerts, W. Sansen, J. Vandewalle, G. Vantrappen, J. Janssens. "Two
methods for optimal mecg elimination and fecg detection from skin electrode signals". IEEE
Trans Biomed Eng March 1987;34(3):233-243.
[11] A. Hyvarinen 1999 "Fast and robust fixed point algorithm for independent component analysis"
IEEE Trans. Neural Netw. 10 626–34.
[12] P. Comon 1994 "Independent component analysis: a new concept" Signal Process. 36 287–314.
[13] J. F. Cardoso 1989 "Source separation using higher order moments" Proc. ICASSP pp 2109–12.
[14] C. Jutten and Taleb "A 2000 Source separation: from dusk till dawn" Proc. ICA2000 pp 15–26.
[15] Geng Jun "Find peak value of datas". USTB, Beijing, China for Dr. Ma Zheng, [Online].
Available: E -mail: dr.gengjun@126.com
[16] (ANSI/AAMI EC57): "Testing and reporting performance results of cardiac rhythm and ST
segment measurement algorithms", 1998. (AAMI Recommended Practice/American National
Standard).

View publication stats